When accessing Ops Manager web dashboard, browser shows "An error occurred. Please contact VMware technical support. Exception: undefined method `access_token’ for nil:NilClass"
This can happen especially after restarting Ops Manager VM.
If one user is on the Apply Changes page (/install
) watching the ongoing Apply Changes when the Ops Manager process was manually restarted by a user on the VM, the javascript on that page that continuously retrieves the logs for display on that page has retry logic when it gets disconnected. It will make a new request to the /api/v0/installations/current_log
API endpoint for the logs. The problem is that if the retry also encounters an error, it will continue retrying over and over, immediately. This floods the Ops Manager application with numerous requests, which exhausted the database connection pool for the application.R&D is adjusting the javascript to add a delay in the retry mechanism, as well as a max retry count. That should help prevent situations like this in the future.
Ask every person that have access to Ops Manager to close the browser tabs accessing Ops Manager and restart tempest-web service (E.gsudo service
tempest-web
restart
). This should solve the issue.